*3.4. Complementation Studies in L. major HSP23-Null Mutants Indicate a Conserved Function in Thermotolerance for Trypanosomatid HSP23*

The failure to establish ectopic HSP23 expression in the *L. braziliensis HSP23*–/– clones precluded a conclusive correlation between loss of HSP23 and the observed phenotypes. To complement this, we also produced CRISPR-derived *L. major HSP23*–/– null mutants, following the same experimental strategies. Three selected *L. major HSP23*–/– null mutant clones (*LmjHSP23*–/– cl.1–cl.3) were analysed by whole genome sequencing, confirming the successful replacement of the *LmjHSP23* gene (Figure S10A) and the correct integration of both drug-resistance cassettes (Figure S11). Further verification by Western blot analysis using HSP23-specific antibodies showed a lack of the HSP23 protein in all *L. major HSP23*–/– null mutants (Figure S10B). From these clones, we selected *LmjHSP23*–/– cl.1 for genetic complementation and phenotypic analyses. We introduced the *LmjHSP23* transgene as episome to generate a *LmjHSP23* add-back cell line. In vitro, at optimal growth conditions for promastigotes (25 ◦C, pH 7.4), the null mutant showed a 50% reduced growth compared with wild-type cells (Figure 5A). This reduced growth of the null mutant could be restored to near-wild type levels by the *LmjHSP23* transgene, but not by the empty expression plasmid pCL1S (Figure 5A). At 34 ◦C, a temperature relevant for dermotropic *Leishmania* species, the *LmjHSP23*–/– cl.1 mutant promastigotes were severely affected and did not proliferate (Figure 5B). This temperature-sensitive phenotype was rescued by the *LmjHSP23* transgene (Figure 5B), similar to what was reported for *L. donovani HSP23*–/– mutants [47]. We also tested the *LmjHSP23*–/– cl.1 mutant for tolerance to sublethal ethanol stress. A 2% ethanol exposure caused growth inhibition in the null mutant (Figure 5C), but not in the *LmjHSP23–*/*–* (LmjHSP23) parasites (Figure 5C). Thus, we established *LmjHSP23*–/– cl.1 as a suitable host strain for the functional complementation with various trypanosomatid *HSP23* genes.

A similar ploidy analysis was also performed for *L. major* WT, *L. major* WT [Cas9] and the two *L. major HSP23–*/*–* clones, 1 and 2 (Figure S9B). Except for a very minor increase for chromosomes 5, 6, and 8, no karyotypic changes could be observed.

The *LmjHSP23*–/– cl.1 mutant was transfected with pCL1S bearing the *L. donovani, L. infantum*, *L. major*, *L. braziliensis*, or *T. brucei HSP23* orthologs, respectively. Ectopic expression of these transgenes was verified at the RNA level using qRT-PCR analysis with *HSP23* species-specific primers, showing varying rates of over expression (Figure S12A). We also verified the HSP23 protein level by Western blot analysis using specific antibodies raised against *L. donovani* HSP23 [47]. Over expression was confirmed for all *Leishmania* HSP23 homologs, except for the putative *T. brucei* HSP23 (Figure S12B), the latter likely due to low amino acid sequence conservation (36%) between the *L. donovani* and *T. brucei* HSP23 homologs.

**Figure 5.** Phenotypic analysis of *L. major HSP23–*/*–* mutants and complementation strains. 1 × 10<sup>6</sup> or 5 × 10<sup>6</sup> parasites/ml were seeded in 10 ml complete M199 medium and parasite density was assessed at day 4. Parasites were grown at 25 ◦C (**A**), 34 ◦C (**B**), and 25 ◦C with 2% EtOH (**C**). Cell density is shown as percentage of WT (set at 100%). (**D**) Complementation studies in *LmjHSP23*–/– mutants. Null mutants were transfected with the pCL1S over expression vector harbouring the *HSP23* gene of *L. major, L. donovani, L. infantum, L. braziliensis*, and *Trypanosoma brucei* or with the empty vector only. Complementation populations were subjected to growth experiments at 34 ◦C. Cell density was assessed at day 4 and is shown normalised to *Lmj* WT growth (set at 100%). \* = *p* < 0.05.

We then tested whether the temperature-sensitive phenotype of the *L. major HSP23–*/*–* mutant could be complemented by the HSP23-encoding, orthologous genes from other *Leishmania* species and the closely related *Trypanosoma brucei*. These supposed HSP23 homologs share between 36% and 99% amino acid sequence identity (Table S2). At 34 ◦C, all trypanosomatid *HSP23* transgenes restored growth of *L. major HSP23*-null mutants to wild-type levels, abrogating the mutant phenotype (Figure 5D). This shows that all trypanosomatid HSP23 homologs share the same functionality, conferring protection against heat stress, and likely maintaining protein folding homeostasis in trypanosomatid organisms. Furthermore, the functional conservation of HSP23 homologs among the Trypanosomatidae confirms the phenotypes we observed in the *L. braziliensis HSP23*-null mutants, since LbrHSP23 expression can restore thermotolerance to the *L. major HSP23*–/– mutant.

#### **4. Discussion**

The protozoan parasite *Leishmania braziliensis* is one of the most pathogenic dermotropic *Leishmania* species circulating in the Americas, where it is the main cause of cutaneous and mucocutaneous leishmaniasis [4,63]. Despite its prevalence and importance to public health, *L. braziliensis* has been less studied and is therefore less experimentally developed compared to Old World *Leishmania* species such as *L. major* and *L. donovani*, which have been traditionally used as models for studying the biology of these obligate intracellular parasites. Given that *L. braziliensis* is a member of the subgenus *Viannia,* with a considerable phylogenetic distance to the Old World species and even to the Central and South American *L. mexicana* complex, conservation of gene function between the subgenera may not be assumed automatically, and may require experimental confirmation by reverse genetics.

One of the main approaches for genetic modification of *Leishmania* parasites to probe gene function has been the generation of gene replacement mutants by homologous recombination-mediated replacement [5,64], which allows the creation of null mutants and their subsequent phenotypic analysis [6,65]. While this has proven a powerful genetic tool in Old World *Leishmania* spp., but also in Central American *L. mexicana* [66], our literature search did not turn up any work regarding homologous recombination-based gene replacement in *L. braziliensis*. Studies reporting on the use of homologous recombination in *L. braziliensis* demonstrate the generation of stable transgenic parasite lines from integration of DNA constructs into the SSU rDNA genomic locus. These include *L. braziliensis* lines expressing reporter genes, e.g., luciferase or eGFP, which hold potential for parasite tracking and monitoring effects of antileishmanial compounds *in vitro* and *in vivo* [67–69], and over expressing parasite lines for the analysis of gene products, e.g., to assess antimony susceptibility and resistance mechanisms [70–72]. Moreover, circular extrachromosomal cosmids can be stably introduced into *L. braziliensis* to over-express stretches of genomic DNA and connect the over expression phenotypes to biological processes such as virulence [73] and antimony resistance [59]. The experimental proof that *L. braziliensis* is a RNAi-competent species started the development of RNAi-based gene knockdown strategies for the loss-of-function phenotyping of genes in this species [9,10]. More recently, the CRISPR–Cas9 technology, with its advantages of being less time-consuming than traditional gene targeting and less susceptible to off-target effects than RNAi-based approaches [74], has added to the genetic toolbox that is available for the study of *Leishmania* spp. [19,20], allowing researchers to investigate gene functions with unprecedented ease, accuracy, efficiency, and scale in biological contexts [17,25,29,40].

In this study, we report the application of CRISPR–Cas9-mediated gene editing to the efficient and precise disruption of two endogenous, non-essential, single-copy genes and one integrated transgene in *L. braziliensis*. We opted for a CRISPR–Cas9, molecular cloning-free method developed for the use in *Leishmania* that relies on T7 RNAP-based expression of sgRNAs in vivo [17]. For this, we first generated a parental *L. braziliensis* cell line expressing Cas9 and T7 RNAP. Since plasmid pTB007 was designed for integration of both transgenes into the *L. major* beta-tubulin locus [17], we transfected pTB007 as stable, circular episome under hygromycin B selection. This episome was well tolerated by *L. braziliensis* strain PER005cl2 used in this study and was stably maintained for several months, with no apparent Cas9 toxicity during in vitro promastigote passage, indicating that this episomal transgene could be maintained without inducing deleterious RNAi effects in *L. braziliensis*.

For our study, we used a cloned *L. braziliensis* strain, derived from a clinical isolate, whose entire genome had been sequenced [46].This allowed us to select correct, highly specific sgRNA templates and donor DNAs for precise, targeted gene editing with no predicted off-target mutations. The original clinical isolate from which PER005cl2 strain is derived, was shown to be infective for primary mouse peritoneal macrophages [34], within which it is sensitive to pentavalent antimony. Furthermore, this isolate was confirmed not to harbour *Leishmaniavirus* LRV1 [75], a cytoplasmic double-stranded RNA virus frequently found as endosymbiont in *Leishmania* (*Viannia*) species [75–77], and which appears to enhance virulence and persistence of its *Leishmania* host [78,79].

We first targeted an eGFP coding sequence inserted into the SSU rRNA coding gene(s) of the *L. braziliensis* parental Cas9/T7 cell line. We applied double antibiotic selection after CRISPR targeting, using increasing antibiotic pressure at two time points, i.e., predetermined minimal effective concentrations of antibiotics at 24 h post-transfection and until transfectants emerged in culture, followed by higher antibiotic selection pressure to enrich for homozygously edited cells, and found this

to be an effective strategy. The *eGFP* editing in *L. braziliensis* was assessed at the cell population level and compared to that achieved in *L. donovani*. Overall, we observed a different activity for the same pairs of sgRNAs in the two *Leishmania* species studied. While all 6 sgRNA sets that targeted sites within the *eGFP* gene were highly active in *L. donovani*, they had a wide range of efficiency in *L. braziliensis*. The most active sgRNAs (sets 5 and 6) were the same in *L. braziliensis* and *L. donovani*, indicating that the sgRNA sequence had an impact on the gene targeting efficiency. This is in line with a recent study that tested the efficiency of three gRNAs targeting identical sequences of the miltefosine transporter gene in *L. donovani*, *L. major*, and *L. mexicana*, and found the relative gRNA activity to be the same [31]. Studies in other systems revealed that sgRNA sequence features such as position-specific nucleotide composition, GC content, motifs located in the sgRNA "seed" region, and secondary structures of sgRNAs contribute to sgRNA efficacy [80–84].

The different gene targeting efficiencies of the same sgRNA sets observed for *L. braziliensis* and *L. donovani* may be due to different factors. First, the presence of an active RNAi machinery in *L. braziliensis* [9] may have an effect on ectopic Cas9 and T7 RNAP expression from episomal DNA constructs in this species, as was shown before [59]. Second, there may be differences in the T7-dependent expression level of different sgRNAs and Cas9 among *Leishmania* species [31]. We have used T7 RNAP-driven in vivo expression of sgRNA templates that were delivered to the *Leishmania* parental Cas9/T7 cell lines by transient transfection [17]. Variation of T7 RNAP-mediated transcription may lead to different intracellular levels of sgRNA that may limit the efficiency of Cas9-dependent DNA cleavage. A recent study suggested that a threshold level for both Cas9 and sgRNA expression is required for an efficient CRISPR-mediated gene knockout, which in turn is determined by the specific potency of a given sgRNA [85]. In keeping with this, increased sgRNA expression and maturation dramatically improved the efficiency of CRISPR–Cas9 mutagenesis in *Candida albicans* [86]. Thirdly, DSB repair efficiency may differ between *Leishmania* species [31]. Fourth, small variations in the intrinsic antibiotic sensitivity of different *Leishmania* species and strains may cause differences in transgene copy numbers, both for the integrated *GFP* gene and for the Cas9/T7-RNAP construct, leading to different efficiencies. Lastly, other factors playing a role in the biology of the *Leishmania* species studied may also play a role, such as variations of chromatin structure.

In our experiments, the copy numbers of *eGFP* within the SSU rRNA gene units of the *L. braziliensis* Cas9/T7/eGFP parental cell line were not determined. Assuming one copy of *eGFP* present per genome in the *L. braziliensis* Cas9/T7/eGFP, as shown in a recent study focused on the same species [87], our results suggest that the *eGFP*-specific sgRNA sets 1, 2, 3, and 4 generated mono-allelic edits, i.e., single-allele replacements, whereas the most efficient sgRNAs, sets 5 and 6, generated mostly double-allelic edits.

We were also able to efficiently disrupt two non-essential, endogenous, single-copy genes of *L. braziliensis* encoding the heat shock proteins HSP23 and HSP100. We obtained double-allelic, Cas9-free *HSP23*–/– and *HSP100*–/– null mutants. The in vitro phenotypes of the *L. braziliensis HSP23* and *HSP100*-null mutants were assessed and compared to the wild-type strain, since gene add-back variants could not be obtained. Nevertheless, the analysis of independently cloned mutant cell lines revealed largely consistent phenotypes, strengthening the correlation between the disruption of the target gene and the loss-of-function phenotypes. This was further supported by the complementation studies carried out in the *L. major HSP23*-null mutant, which demonstrated functional homology between the *HSP23* genes of the Trypanosomatidae. Furthermore, the rapid loss of the Cas9 episome in the absence of antibiotic selection is important when evaluating the phenotype, as the WT [Cas9] strain which was kept under selection showed a divergent phenotype from the wild type. We would therefore refrain from using genomic integration constructs for the expression of Cas9.

We do not know the reason behind the different capacity of intracellular amastigotes from the three studied *L. braziliensis HSP23*–/– mutants to survive inside macrophages. All parasite strains/clones were subjected to the same *in vitro* culture, electroporation, cloning, antibiotic selection, and stress conditions. They had similar passage numbers before phenotype analyses, and their phenotypes were investigated in parallel in all assays. Moreover, the CRISPR–Cas9 components were no longer present when single-cell cloning was performed. We suspected that the mutant clones might have undergone some level of genetic adaptation, e.g., via spontaneous mosaic aneuploidy followed by selection for vitality. We observed a similar, spontaneous loss of phenotype for a *L. donovani HSP23*–/– clone, due to amplification of the gene coding for casein kinase 1.2 [45]. We indeed found ploidy changes that were specific to the *L. braziliensis HSP23*–/– mutants. One of those, a trisomy of chromosome 34, which harbours the casein kinase 1.2 gene in *L. braziliensis*, may have a similar effect as in *L. donovani*.

Lastly, an average of 37.7% (± 8.6%) *L. braziliensis* Cas9-expressing cells were able to survive inside macrophages (Figure 4E). Those cells show a trisomy for chromosome 26, similar to all three *L. braziliensis HSP23*–/– clones (Figure S9A). This trisomy is absent from the wild type and from the two *L. braziliensis HSP100*–/– clones.
